astro-legacy-archive/fermi-lat-weekly-photons

Fermi-LAT weekly photons

This dataset contains Fermi Large Area Telescope all-sky weekly photon files from mission week w009 through w153, frozen on 2026-08-30. Its 145 configurations correspond one-to-one with the weekly p305_v001 FITS files. Each Parquet row is an EVENTS row, with the 23 FITS-named columns in their stored order and shape.

Mission weeks run Thursday through Wednesday in UTC. The first configuration begins with the science-phase interval on 2008-08-04; w153 ends on 2011-05-12. FSSC states: “The current mission week file is regenerated by the FSSC after every data delivery from the LAT instrument team.”

Configurations	w009–w153 (145 files)
Photon events	228,828,995 rows
Snapshot date	2026-08-30
Source processing/version	p305_v001
Source FITS volume	22,429,635,840 bytes
Sky coordinates	FK5 J2000 (RA, DEC) and Galactic (L, B)

How to use

python -m venv .venv && .venv/bin/pip install datasets huggingface_hub pyarrow

from datasets import load_dataset
dataset = load_dataset("astro-legacy-archive/fermi-lat-weekly-photons", "w009", split="train[:1]", columns=["ENERGY", "L", "B"])
row = dataset[0]
print(row["ENERGY"], row["L"], row["B"])

44.50176239013672 123.25150299072266 11.777127265930176

import json
from huggingface_hub import hf_hub_download
import pyarrow.parquet as pq
path = hf_hub_download("astro-legacy-archive/fermi-lat-weekly-photons", "w009.parquet", repo_type="dataset")
print(json.loads(pq.read_metadata(path).metadata[b"event_list"])["schema"])

astro-legacy-archive/event-list

The embedded event-list document records the source HDU, field definitions, units, and bit-array representation. EVENT_CLASS and EVENT_TYPE mirror each FITS 32X value as 32 Boolean elements in stored order: for a published integer mask M, stored element i corresponds to integer bit 31 - i. The same relation applies to the published conversion, PSF, and energy-dispersion masks.

Dataset structure

column	type	unit	meaning
ENERGY	float32	MeV	Reconstructed event energy
RA, DEC	float32	deg	Reconstructed FK5 J2000 direction
L, B	float32	deg	Reconstructed Galactic direction
THETA, PHI	float32	deg	Incidence direction relative to the LAT boresight and spacecraft +X axis
ZENITH_ANGLE, EARTH_AZIMUTH_ANGLE	float32	deg	Reconstructed direction in local Earth coordinates
TIME	float64	s	Mission elapsed time
EVENT_ID, RUN_ID	int32	—	Event sequence number and LAT data-acquisition-period identifier
RECON_VERSION	int16	—	Event-reconstruction software version
CALIB_VERSION	list[int16, 3]	—	Calibration-table versions for the anticoincidence detector, calorimeter, and tracker; FSSC marks this column "currently unused"
EVENT_CLASS, EVENT_TYPE	list[bool, 32]	—	FITS 32X classifications in stored order
CONVERSION_TYPE	int16	—	Tracker conversion section: front=0, back=1
LIVETIME	float64	s	Short-term accumulated LAT livetime
DIFRSP0–DIFRSP4	float32	—	Diffuse-response components

Mission week w009 observations

The w009 table is time-ordered with no descending TIME steps. It contains 88,612 front-converting and 82,918 back-converting events. EVENT_CLASS has 50 distinct stored masks and EVENT_TYPE has 32. The standard hierarchy nests SOURCE, CLEAN, ULTRACLEAN, and ULTRACLEANVETO; SOURCEVETO is nested in SOURCE but not in the other three. The public integer masks are P8R3_SOURCE=128, P8R3_CLEAN=256, P8R3_ULTRACLEAN=512, P8R3_ULTRACLEANVETO=1024, and P8R3_SOURCEVETO=2048.

All five DIFRSP0–DIFRSP4 columns contain exactly 171,530 zero values. FSSC documents standard Pass 8 files as released without populated diffuse responses; weekly files with precomputed diffuse columns are a separate product.

Mission week 009 is shorter than the normal Thursday–Wednesday interval. The source header spans 2008-08-04T15:43:36 through 2008-08-07T00:04:27.9999 and begins with the first science-phase interval; FSSC dates the start of science operations to 2008-08-04. The source headers retain the pre-launch telescope name TELESCOP='GLAST'; the instrument is INSTRUME='LAT'.

FSSC distinguishes the Photon File used here from the Extended File, which contains looser transient-class events and more event-level quantities, and the Spacecraft File, which records position, orientation, and observing efficiency. Analyses requiring spacecraft history or the extended event selection need those companion products.

Provenance

The source census is recorded in sources.json, with the configuration, filename, URL, byte size, SHA-256 digest, and snapshot date for every published weekly file. Each Parquet file embeds the corresponding source identity and digest.

Every Parquet file carries a versioned event_list metadata document with its complete primary, EVENTS, and GTI headers; all source GTI rows; field descriptions, units, and FITS definitions; classification bit mapping; source URL and hash; credit; and regeneration command. PyArrow schema inspection exposes field descriptions and units; load_dataset retains types but strips field metadata. The published table is the source EVENTS bintable. The source GTI bintable's START and STOP arrays are carried in full in the embedded metadata and can be read alongside the event table.

Licensing and credit

No explicit data licence is published on the FSSC data-access, product, or data-policy pages. HEASARC states:

HEASARC materials are all available freely for your use.

This is a free-use statement, not a named or formal licence. HEASARC requests:

This research has made use of data and/or software provided by the High Energy Astrophysics Science Archive Research Center (HEASARC), which is a service of the Astrophysics Science Division at NASA/GSFC.

Citations

Please cite the LAT instrument paper, the on-orbit event-classification and calibration paper, the Pass 8 revision, and the P8R3 selection release.

Errata and requests

To report an erratum, propose a correction, or ask a question about this dataset, open a Discussion or Pull Request in its Community tab; posting requires a free Hugging Face account. Dataset requests and archive-wide questions may be posted in the Community tab of any dataset. Confirmed corrections are made in the source conversion tools and published by rebuilding the dataset.